| The change of the Canopy Chlorophyll Content(CCC)plays an important role in plant disasters monitoring and ecosystem health."Ruo Qiang grey jujube" planting in Xinjiang known as the " Gourmet Jujube".it is significant to estimate the canopy chlorophyll content of jujube more accurately and quickly for grasping the growth state and estimating the yield of jujube in time.At present,there are few studies on the estimation of chlorophyll content in the canopy of jujube in Xinjiang and the remote sensing data sources used in the estimation process are relatively single,it is difficult to achieve a larger range of monitoring and can not reflect the overall vegetation status using single data source.In terms of methods,traditional studies use simple linear regression models or general machine learning methods,doesn’t consider the possible influence of different planting locations of vegetation on the chlorophyll content of vegetation.In view of this,this study Extracted gray jujube area based on object-oriented + canopy hyperspectral + Sentinel 2 image analysis +SVM classification algorithm,The spatial aggregation of chlorophyll content in canopy was analyzed.The feasibility of GWLS-SVR(geographically weighted least support vector regression)model was discussed,and the optimal vegetation index parameters and GWLS-SVR model were used to estimate the chlorophyll content in the canopy of Jujube.The results were as follows:(1)Gray jujube was the main type of jujube planted in the study area,the planting area of gray jujube studied in this paper is 35.27 km2,and the distribution is relatively concentrated,showing a zoned distribution within the scope of the study area.(2)Moran’s I index of chlorophyll content in canopy indicating that the chlorophyll content in canopy at two growth stages of two data sources was very suitable for GWLS-SVR modeling.in this study,GWLS-SVR model reflects its special advantage,based on the flowering period and mature period and the precision of the model in different data source were higher than MLR and SVR model,and four modeling R2 were higher than 0.89,the MSE is below 1.9,at the same time,the model can make up for the correlation of vegetation index and low canopy chlorophyll content,improve the accuracy of inversion,has the very good universality.(3)By screening vegetation index modeling parameters,we found that MCARI with one red edge band and vegetation indices such as MTCI with 2 red edge bands had a high correlation with chlorophyll content in canopy of Jujube at flowering stage.However,this situation changed in the mature stage in September.The correlation between DVI and NPCI without red edge band and canopy chlorophyll content was relatively high in the mature stage,indicating that the correlation between the improvement of vegetation index with red edge band and canopy chlorophyll content might be related to the growth stage.(4)The dynamic analysis of the estimated distribution map of canopy chlorophyll content from different data sources and growth stage,it was found that the spatial distribution of canopy chlorophyll content estimated from two data sources at flowering and maturity stage was roughly similar.but there are some differences in the numerical distribution,The canopy chlorophyll content collected by Sentinel-2 data is more average,and the estimated results are more refined compared with the hyperspectral data.Through the dynamic analysis of the four results estimated by two data sources in the two growth stages of flowering and maturity,it was found that the canopy chlorophyll content of jujube had a great increase from flowering to maturity,and the growth condition of jujube with low value in flowering period had a great improvement after reaching maturity. |
PDF Full Text Request